Back to EveryPatent.com
| United States Patent |
6,016,993
|
|
Maino
, et al.
|
January 25, 2000
|
Method of monitoring a transmission assembly of a vehicle equipped with
acceleration sensors, in particular a helicopter
Abstract
The method includes the steps of: acquiring a signal from an acceleration
sensor; calculating the Hilbert transform of the signal filtered and
sampled beforehand; defining a complex signal having the filtered and
sampled signal as the real part and the Hilbert transform of the filtered
and sampled signal as the imaginary part; calculating a phase signal given
by the difference between the phase of the complex signal and a reference
phase; calculating the variance of the phase signal; comparing the
variance with at least one predetermined threshold; and generating an
alarm signal if the variance exceeds the threshold value.
| Inventors:
|
Maino; Bruno (Samarate, IT);
Bellazzi; Alberto (Cameri, IT)
|
| Assignee:
|
Finmeccanica S.p.A. (Rome, IT)
|
| Appl. No.:
|
110305 |
| Filed:
|
July 6, 1998 |
Foreign Application Priority Data
| Jul 04, 1997[IT] | TO97A0592 |
| Current U.S. Class: |
244/17.13; 180/337; 180/338; 244/60 |
| Intern'l Class: |
B64C 011/34 |
| Field of Search: |
244/60,17.13
180/337,338
|
References Cited
U.S. Patent Documents
| 3699806 | Oct., 1972 | Weichbrodt | 73/71.
|
| 4654836 | Mar., 1987 | Wason | 367/190.
|
| 5210704 | May., 1993 | Husseiny | 364/551.
|
| 5344101 | Sep., 1994 | Francois | 244/60.
|
| 5365787 | Nov., 1994 | Hernandez et al. | 73/660.
|
| Foreign Patent Documents |
| 2 234 411 | Jan., 1991 | GB.
| |
| WO 96/05486 | Feb., 1996 | WO.
| |
Primary Examiner: Eldred; J. Woodrow
Attorney, Agent or Firm: Darby & Darby
Claims
We claim:
1. A method of monitoring a transmission assembly of a vehicle equipped
with acceleration sensors, comprising the steps of:
a) acquiring a signal from an acceleration sensor mounted on the
transmission assembly for detecting the acceleration of a moving member of
the transmission assembly;
b) calculating the Hilbert transform of said signal;
c) defining a complex signal having said signal as the real part and said
Hilbert transform of said signal as the imaginary part;
d) calculating the difference between the phase of said complex signal and
a reference phase to obtain a phase signal;
e) calculating the variability of said phase signal to obtain a variability
signal;
f) comparing said variability signal with at least a first predetermined
threshold; and
g) generating an alarm signal if said variability signal exceeds said at
least a first predetermined threshold.
2. A method as claimed in claim 1, characterized in that said variability
signal is the variance of said phase signal.
3. A method as claimed in claim 1, characterized by repeating said steps a)
to f) for a number of iterations.
4. A method as claimed in claim 1, characterized in that said step of
calculating said Hilbert transform is preceded by a step of sampling and
filtering said signal.
5. A method as claimed in claim 4, characterized in that said step of
sampling and filtering comprises the steps of:
acquiring a number of synchronous samples T.sub.1 (i) for each revolution
of a shaft associated with said acceleration sensor; and
calculating an average time series T.sub.1m (i) according to the equation:
##EQU5##
where T.sub.1 (i) is said synchronous samples, and 1 is a counter for
counting the number of revolutions of said shaft.
6. A method as claimed in claim 5, characterized in that said step of
sampling and filtering also comprises the step of filtering said average
time series (T.sub.1m (i)) with a band-pass filter centered at a
predetermined frequency, to obtain a filtered mean signal (T'.sub.1m (i)).
7. A method as claimed in claim 5, characterized in that said step of
calculating an average time series is preceded by a step of determining
the presence of significant signal values.
8. A method as claimed in claim 5, characterized in that said step of
calculating an average time series is followed by a step of determining
convergence of said average time series.
9. A method as claimed in claim 1, characterized in that said step of
comparing said variability signal also comprises the step of comparing
said variability signal with a second threshold.
Description
BACKGROUND OF THE INVENTION
The present invention relates to a method of monitoring a transmission
assembly of a vehicle equipped with acceleration sensors, in particular a
helicopter.
As is known, helicopter transmission members must be monitored continuously
to immediately detect any faults or incipient malfunction conditions, and
so prevent failure during flight.
For this purpose, the transmission is equipped with acceleration sensors,
the signals of which are processed to determine any faults on the
transmission. The methods currently used, however, to process the fault
detection sensor signals are not sensitive enough to ensure fault
detection well in advance of catastrophic failure.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a monitoring method
ensuring reliable, advance detection of any faults or malfunction
conditions which might result in failure.
According to the present invention, there is provided a method of
monitoring a transmission assembly of a vehicle equipped with acceleration
sensors, in particular a helicopter, characterized by comprising the steps
of:
a) acquiring a signal from an acceleration sensor;
b) calculating the Hilbert transform of said signal;
c) defining a complex signal having said signal as the real part and said
Hilbert transform of said signal as the imaginary part;
d) calculating the difference between the phase of said complex signal and
a reference phase to obtain a phase signal;
e) calculating the variability of said phase signal to obtain a variability
signal; and
f) comparing said variability signal with at least a first predetermined
threshold.
BRIEF DESCRIPTION OF THE DRAWINGS
A number of non-limiting embodiments of the invention will be described by
way of example with reference to the accompanying drawings, in which:
FIG. 1 shows a top plan view of a helicopter, in which the helicopter
transmission is shown schematically;
FIG. 2 shows a simplified diagram of the helicopter transmission and the
location of various transmission member sensors;
FIGS. 3-5 show flow charts of steps in the method according to the
invention.
DETAILED DESCRIPTION OF THE INVENTION
FIGS. 1 and 2 show, schematically, the members of a helicopter 100
pertinent to the present invention.
In particular, helicopter 100 comprises a first engine 101, a second engine
102, and a third engine 103; first engine 101 is connected, by means of a
first transmission line 104 comprising a first, second and third reducer
105-107, to an input gear 108 of a main gear assembly 109; second engine
102 is connected to input gear 108 by means of a second transmission line
111 comprising a fourth, fifth and sixth reducer 112-114; and third engine
103 is connected to input gear 108 by means of a third transmission line
116 comprising a seventh, eighth and ninth reducer 117-119.
Input gear 108 is connected to a rotor 121 of helicopter 100 by means of an
epicyclic reducer 122 forming part of main gear assembly 109 and
comprising six planet gears 123 and a sun gear 124; and input gear 108 is
also connected--by means of a fourth transmission line 125 also connected
to first transmission line 104--to an accessory box indicated
schematically by 126, and to a fifth transmission line 130 connected to a
tail rotor 134 and comprising a power take-off gear 131, intermediate
joints 135-137, an intermediate gearbox 139, and a tail gearbox 140.
FIG. 2 shows fifteen acceleration sensors 1-15 and two azimuth sensors 16,
17 fitted close to the reduction stages as indicated in Table I below.
TABLE I
______________________________________
Sensor Pos. Sensor Pos. Sensor
Pos.
______________________________________
1 105 7 131 13 140
2 112 8 122 (front)
14 136
3 117 9 122 (right)
15 137
4 107 10 122 (left)
16 108
5 114 11 126 17 139
6 119 12 139
______________________________________
Helicopter 100 also comprises a data processing unit 150 connected to
sensors 1-17 by an interface unit 151 for sampling and digitizing the
sensor signals, and to a data memory 152, an event memory 153, and a
mathematical processor 154.
The monitoring method described below provides, by analyzing the signals
from acceleration sensors 1-7, 12 and 13, for detecting problems arising
in flight, in particular, mechanical problems, torque transmission
problems, structural problems involving cracks, and signal phase
modulation problems; in all of which cases, at least one of the analyzed
signals comprises a peak which is detectable by the method described.
The monitoring method provides for processing a signal s(t), supplied by
whichever of acceleration sensors 1-7, 12, 13 is considered in each case,
using a signal s1(t) supplied by the azimuth sensor on a shaft connected
to the reducer monitored by the sensor in question (azimuth sensor 16 for
acceleration sensors 1-7, and azimuth sensor 17 for acceleration sensors
12, 13). Signal s(t) is a vibratory signal (related to the rotation
frequency of the shaft associated with the sensor in question) and is
affected by random noise as well as by noise related to other nearby
rotary members.
More specifically, and with reference to FIG. 3, the method comprises an
initial step, in which a variable F (explained later on with reference to
FIG. 5) is set to a first predetermined value, e.g. 0 (block 20).
Signal s(t) and signal s1(t) are then acquired (block 21), and a sampling
frequency FS is calculated by multiplying the frequency of azimuth sensor
signal s1(t) by a memorized coefficient KT equal to the ratio between two
prime whole numbers, and in particular correlated to the transmission
ratio between the azimuth sensor shaft and the shaft monitored by the
sensor in question, so that the resulting sampling frequency FS is
correlated to the azimuth sensor frequency, and is such as to supply
exactly NJ points (where NJ is a power of 2) of signal s(t) for each
revolution of the monitored shaft.
Signal s(t) is then sampled at frequency FS and filtered by interface unit
151 to remove the random noise and nonsynchronous periodic components
(block 22) and obtain a filtered signal T.sub.1m (i) defined by a number
of samples indicated by "i". Signal s(t) is preferably sampled and
filtered using the sequence of steps shown in FIG. 4 and described later
on.
Signal T.sub.1m (i) is further filtered using a band-pass filter centered
on the gear meshing frequency, to obtain a filtered mean signal T'.sub.1m
(i) (block 24).
The Hilbert transform H[T'.sub.1m (i)] of signal T'.sub.1m (i) is then
calculated (block 25); and a complex signal C.sub.1m (i) related to signal
T'.sub.1m (i) is calculated (block 26) according to the equation:
C.sub.1m (i)=T'.sub.1m (i)+jH[T'.sub.1m (i)]
The initial phase P.sub.0 of signal C.sub.1m (i) is then determined and
memorized (block 27); and instantaneous phases P(i) are determined and
memorized (block 28) according to the equation:
##EQU1##
where Z is the number of gear teeth.
The phase FC(i) of complex signal C.sub.1m (i) is then calculated (block
29) according to the equation:
FC(i)=arg[C.sub.1m (i)]-P(i)-P.sub.0
where arg[C.sub.1m (i)] is the argument of complex number C.sub.1m (i),
defined as arctg{T'.sub.1m (i)/H[T'.sub.1m (i)]}, and P(i) and P.sub.0 are
the previously calculated phase values.
The variance V of signal FC (i) is then calculated (block 30); the
resulting value V is compared by central processing unit 150 with two
threshold values TH1, TH2 (where TH1<TH2) to generate, if necessary, alarm
signals (block 31) according to the procedure described in detail later on
with reference to FIG. 5; and block 31 then goes back to block 21 to
continue monitoring with the next portion of signal s(t).
Sampling and filtering in block 22 are conveniently performed using the
method described below with reference to FIG. 4.
To begin with, a revolution counter 1 is set to 1 (block 41); and signal
s(t) is sampled at the previously defined frequency FS to obtain NJ points
or samples T.sub.1 (i) representing a synchronous vibratory time series
relative to the sensor in question, to the respective shaft, and to each
1-th revolution (block 42).
The value of counter 1, in particular whether it is less than or equal to a
predetermined value K1, is determined (block 43); and, if counter 1 is
less than or equal to K1 (performance of fewer than K1 iterations
corresponding to the revolutions of the monitored shaft--YES output of
block 43), the availability of the signal is determined by calculating
(block 47) the sum .DELTA..sub.d of the samples acquired at the 1-th
revolution, according to the equation:
##EQU2##
A limit value .DELTA..sub.1 is then calculated (block 48) according to the
equation:
.DELTA..sub.1 =K2.times.FSV.times.NJ
where FSV is the bottom-scale value, and K2 a predetermined constant much
lower than 1; and a check is made (block 49) to determine whether the
calculated sum .DELTA.d is less than the limit value .DELTA..sub.1. In the
event of a negative response (NO output of block 49), the signal is
considered to exist, counter 1 is increased (block 50), and block 50 goes
back to block 42 to acquire further NJ points relative to the next
revolution. Conversely (YES output of block 49), the signal is considered
nonexistent and an alarm signal is generated (block 51); a check is made
(block 52) to determine the presence of a substitute sensor (e.g. sensor 6
for sensors 4 and 5); in the event of a positive response (YES output of
block 52), block 52 goes back to block 21 in FIG. 3 to repeat the
procedure on the substitute sensor; and, in the absence of a substitute
sensor, or if convergence is not reached even with the substitute sensor
(NO output of block 52), the procedure is interrupted.
Conversely, if counter 1 is greater than K1 (performance of more than K1
iterations corresponding to the revolutions of the monitored shaft--NO
output of block 43), block 43 goes on to a block 55, which calculates the
average contiguous synchronous time series T.sub.1m (i) defining the
filtered signal calculated in block 22, according to the equation:
##EQU3##
i.e. the mean value of each sample T.sub.1 (i) over the 1 revolutions
considered is calculated.
Convergence of the averaging process is then determined by calculating a
convergence value .DELTA. given by the sum, over all the samples, of the
absolute value of the difference between the actual mean value of each
sample and the mean value calculated in a previous revolution (at distance
4) divided by the sum of the mean samples in the previous revolution
considered (at distance 4), according to the equation:
##EQU4##
where T.sub.1m (i) represents the i-th sample of the 1-th iteration, and
T.sub.1m-4 (i) represents the i-th sample of the 1-4-th iteration (block
56).
A check is then made to determine whether the calculated convergence value
is less than or equal to a predetermined permissible minimum convergence
value .DELTA..sub.C (block 57). In the event of a positive response (YES
output), the convergence process is interrupted, and block 57 goes back to
the main program (block 23 in FIG. 3). Conversely, a check is made to
determine whether the averaging process has already been performed a
predetermined maximum number of times L (block 59). If the iteration
(revolution) counter 1 is less than L (NO output), the counter is
increased (block 50) and the operations described above are repeated.
Conversely (YES output), the procedure for generating an alarm signal and
possibly repeating the procedure with a substitute sensor, as described
with reference to blocks 51-52, is repeated.
The threshold comparison and alarm generating step in block 31 of FIG. 3 is
conveniently performed as described below with reference to FIG. 5.
To begin with, V is compared with first threshold TH1 (block 60); if V<TH1
(YES output), block 60 goes back to block 21 in FIG. 3 to continue
monitoring with the next group of samples; conversely, if threshold TH1
has been exceeded (NO output of block 60), the event is memorized in event
memory 153 (block 62) and V is compared with second threshold TH2 (block
64). If V<TH2 (YES output of block 64), variable F (initialized in block
20 of FIG. 3 to memorize whether threshold TH2 has already been exceeded)
is set to (or confirmed at) the first predetermined value, 0 in the
example shown (block 66), and block 66 goes back to block 21 of FIG. 3 to
continue monitoring with the next group of samples. Conversely (NO output
of block 64), the event is memorized in event memory 153 (block 68) and
the value of variable F is determined (block 70). In particular, if
variable F is at the first predetermined value (YES output), F is set to a
second predetermined value, e.g. 1, (block 72), and block 72 goes back to
block 21 of FIG. 3 to continue monitoring with the next group of samples.
Conversely, if variable F is at the second predetermined value, i.e.
threshold TH2 has already been exceeded (NO output of block 70), a pilot
alarm signal is generated (block 74), variable F is again set to the first
predetermined value (block 76), and block 76 goes back to block 21 of FIG.
3 to continue monitoring with the next samples.
The advantages of the method described are as follows. In particular,
simultaneously controlling the various sensors installed provides for
monitoring different structural parts of the helicopter instant by
instant, and for detecting incipient mechanical problems caused by wear of
the rotary parts of the transmission. Again at a structural level, the
method described provides for detecting cracks in severely stressed parts
of the helicopter, and which may result in rapid failure.
Finally, at signal processing level, the method provides for detecting any
problems arising in modulation of the sensor signals.
Clearly, changes may be made to the method as described and illustrated
herein without, however, departing from the scope of the present
invention. In particular, the filtering and averaging procedure and the
alarm generating procedure may differ from those described.
Top